Method and apparatus for motion vector field encoding

Abstract

A method and apparatus for representing motion in a sequence of digitized images derives a dense motion vector field and vector quantizes the motion vector field.

1. A method of representing motion in a sequence of digitized images comprising generating and coding a plurality of versions of a motion vector field at different resolutions, comprising the steps of:
 producing a motion vector field at a first, coarse, resolution and at a second, finer, resolution; 
 up-sampling the coarse resolution version of the motion vector field; 
 comparing the up-sampled coarse resolution version of the motion vector field with the finer resolution version of the motion vector field; 
 coding the coarse resolution version of the motion vector field; and 
 coding the finer resolution version of the motion vector field on the basis of the results of the comparison. 
 
   
   
     2. A method as claimed in  claim 1  wherein:
 the step of producing a motion vector field at a first, coarse, resolution and at a second, finer, resolution comprises sub-sampling the motion vector field to produce sub-sampled versions of the field at a first, coarse, resolution and at a second, finer, resolution; the step of comparing the coarse resolution field with the finer resolution field produces a residual error, and 
 the step of coding comprises coding the residual error at the finer resolution. 
 
   
   
     3. A method as claimed in  claim 2  comprising producing and coding a sequence of residual errors at increasing resolutions. 
   
   
     4. A method as claimed in  claim 1  comprising multiplexing the coded motion vector field versions at different resolutions into an embedded bit stream. 
   
   
     5. A method as claimed in  claim 1  wherein said coding is performed using vector quantization. 
   
   
     6. An encoder for encoding motion information for a sequence of digitized images according to a method as claimed in  claim 1 . 
   
   
     7. A decoder for decoding motion information for a sequence of digitized images encoded according to a method as claimed in  claim 1 . 
   
   
     8. A method as claimed in  claim 1  wherein at least one of the resolutions corresponds to a dense motion vector field. 
   
   
     9. Apparatus for representing motion in a sequence of digitized images comprising means for generating and coding a plurality of versions of a motion vector field at different resolutions, the apparatus further comprising:
 comprising the steps of: 
 means for producing a motion vector field at a first, coarse, resolution and at a second, finer, resolution; 
 means for up-sampling the coarse resolution version of the motion vector field; 
 means for comparing the up-sampled coarse resolution version of the motion vector field with the finer resolution version of the motion vector field; 
 means for coding the coarse resolution version of the motion vector field; and 
 means for coding the finer resolution version of the motion vector field on the basis of the results of the comparison. 
 
   
   
     10. Apparatus as claimed in  claim 9  comprising means for sub-sampling the motion vector field to produce sub-sampled versions of the field at a first, coarse, resolution and at a second, finer, resolution, means for comparing the coarse resolution field with the finer resolution field to produce a residual error, and means for coding the coarse resolution motion vector field and the residual error at the finer resolution. 
   
   
     11. Apparatus as claimed in  claim 10  comprising means for producing and coding a sequence of residual errors at increasing resolutions.